Pellet Reduction Properties Under Different BF Operating Conditions

8/11/2019 Pellet Reduction Properties Under Different BF Operating Conditions

1/124

LICENTIATE T H E S I S

Lule University of Technology

Department of Chemical Engineering and Geosciences

Division of Process Metallurgy

2006:67|: 02-757|: -c -- 06 67 --

2006:67

Pellet Reduction Properties

under Different Blast Furnace

Operating Conditions

Ulrika Leimalm

Welcome Publications


2/124



3/124

Pellet Reduction Properties under Different Blast Furnace

Operating Conditions

Ulrika Leimalm

Licentiate Thesis

Lule University of TechnologyDepartment of Chemical Engineering and Geosciences

Division of Process MetallurgySE-971 87 Lule

Sweden

2006



4/124



5/124

I

ACKNOWLEDGEMENTSFirst and foremost, I wish to express my gratitude to my supervisors, Professor Bo

Bjrkman and Dr. Lena Sundqvist kvist, for all help, support and supervisionduring my work.

Financial support from the Swedish National Energy Administration (STEM) andthe Agricola Research Centre is gratefully acknowledged.

Thanks to the members of the Swedish Steel Producers Association JK21059project, Closed loop styrning av masugn, for many discussions about the blastfurnace and valuable feedback on my work.

I wish to thank LKAB and SSAB Tunnplt AB for providing materials, tests,analyses and for supporting the work done. A special thanks to LKAB for

providing the opportunity to carry out tests in the LKAB Experimental BlastFurnace.

I am most grateful to Johan Folkesson for his voluntary work in designing thelaboratory furnace control system at LTU and for his continuous support wheneverhardware or software has caused trouble. This work would not have been assuccessful without his help.

I would like to thank Anita Wedholm for all her assistance in the lab and for herpositive attitude. Thanks to my former office mate Fredrik Engstrm and DanielAdolfsson for many enjoyable conversations over coffee breaks and lunches.

Computers seem to live a life of their own, and sometimes unexpectedly refuse tocooperate. Thanks to Bertil Plsson for always putting them on the right track.

I wish to thank my friends and colleagues at the department of ChemicalEngineering and Geosciences. Thanks to my friends outside the academic world forall your support. A special thanks to all of you in the south who, despite thenortherly latitude and sometimes cold climate, came to visit in Lule!

Finally, I would like to thank my parents, my sister and my grandmother.



6/124

II



7/124

III

ABSTRACTOne of the aims of modern blast furnace (BF) ironmaking is to reduce coke

consumption. One way is to increase the injection of reduction agents, such aspulverized coal. An increase in pulverized coal injection rate (PCR) will affect theblast furnace process and the conditions for iron oxide reduction. Changes in PCRinfluence the composition of the ascending gases and the in-furnace temperatureisotherms.

The performed tests involve full-scale, pilot and laboratory investigations.

Raw material sampling of, among other things, pellets was carried out during a

period of fluctuations in the hot metal Si content at the SSAB BF No. 3 at Lule.Although differences in pellet low-temperature reduction disintegration and thehigh-temperature breakdown were observed, the reduction behaviours during blastfurnace simulation tests were almost identical. Differences in the hot metal Sicontent in a production blast furnace were difficult to correlate to raw material

properties, since the process conditions were changed in order to control the heatlevel of the blast furnace.

Tests in the LKAB Experimental Blast Furnace (EBF) were carried out underdifferent pre-set process conditions. Injection of high-volatile (HV) coal resulted ina higher reduction potential in the ascending gas due to a higher H 2content and anincreased shaft temperature compared to operation with low-volatile (LV) coal. Ahigher pellet reduction degree was attained in pellets taken out with the upper shaft

probe during operation with the HV coal compared to injection of the LV coal. Thedifferences receded through the shaft and no differences in pellet reduction degreethat can be correlated to the pre-set process conditions were observed in samplestaken out with the lower shaft probe. However, differences in the pellet texture

were observed. For the HV coal, a higher pellet strength but also an increase ingeneration of Femet fines, was observed compared to operation with the LV coal.Different Femet textures were observed in the pellet depending on the choice ofinjection coal type. The pore size increased and the Femetareas became smootherduring HV coal operation compared to during LV coal operation. The presentresults indicate that it was likely that the Femet texture in the pellet peripheryinfluenced the generation of Femet fines, which left the EBF with the top gas.Further study on the subject is suggested. The present investigations showed an



8/124

IV

increased carburization of Femetand an increase in the K content in pellets taken outwith the lower shaft probe during injection of the HV coal.

Blast furnace simulation laboratory reduction tests for hypothetical PCR indicatedthat an increase in hypothetical PCR was necessary to compensate for the decreasein reduction time between a slow and a fast temperature profile. The reduction timeinfluenced the Femettexture in the pellet periphery.

Blast furnace simulation laboratory reduction for simulated PCR based onmeasurements in the EBF showed that larger pores were observed in the Femet

pellet periphery at high PCR. At simulated low PCR the Femetwas denser. A graintexture was observed in the pellet core after the simulated low PCR, a phenomenon

not found in pellets from the simulated high PCR tests. The present resultsindicated that the texture differences were introduced in the beginning of thereduction.

Results from the EBF tests and laboratory reduction experiments implied that highH2levels in the reduction gas, high heating rates and temperature levels were therequirements for formation of a pellet periphery with large pores.



9/124

V

LIST OF PAPERSThis thesis is based on the results reported in the following papers, which are given

in Appendix I-III.

I. U. Leimalm, L. Sundqvist kvist, A. Brnnmark and B. Bjrkman.Correlation Between Pellet Reduction and Some Blast Furnace OperationParameters

AISTech 2006 Proceedings of the Iron & Steel Technology Conference

U. Leimalm has in collaboration with A. Brnnmark planned and carried out

the experiments related to the EBF test. The major part of the evaluation ofthe results and laboratory reduction experiments has been carried out by U.

Leimalm. Other co-authors have contributed in a supervisory capacity.

II. U. Leimalm, L. Sundqvist kvist and B. Bjrkman. Effect of SimulatedPCI Rate on Olivine Pellet Reduction

Accepted for publication in proceedings from The 4thInternational Congresson the Science and Technology of Ironmaking

Co-authors have contributed in a supervisory capacity.

III. U. Leimalm, L. Sundqvist kvist and B. Bjrkman. Effect of Different PCIPractice on the Texture Obtained During Reduction of Iron Oxide Pellets

Manuscript

Co-authors have contributed in a supervisory capacity.



10/124


11/124

VII

CONTENTS

1 INTRODUCTION...................................................................................................... 1

1.1 BLAST FURNACE IRONMAKING ........................................................................... 11.2 CONDITIONS FOR IRON OXIDE REDUCTION .........................................................61.3 OBJECTIVES......................................................................................................... 9

2 MATERIALS AND METHODS............................................................................. 11

2.1 MATERIALS ....................................................................................................... 112.2 SSAB BLAST FURNACENO. 3 ........................................................................... 112.3 LKAB EXPERIMENTAL BLAST FURNACE ..........................................................122.4 LABORATORY REDUCTION FURNACE ................................................................ 142.5 EVALUATION OF PELLET PROPERTIES ............................................................... 212.6 CHEMICAL ANALYSIS........................................................................................ 22

3 RESULTS ................................................................................................................. 23

3.1 EVALUATION OF PROCESS DATA AND PELLET SAMPLES FROM SSAB BLASTFURNACENO. 3 , APPENDIX I............................................................................. 23

3.2 LABORATORY REDUCTION FOR HYPOTHETICAL PCR, APPENDIX I ...................263.3 EBF TESTS, APPENDIX I-III...............................................................................28

4 DISCUSSION ...........................................................................................................45

4.1 R EDUCTION ....................................................................................................... 454.2 TEXTURE ........................................................................................................... 464.3 PROCESS............................................................................................................ 474.4 GENERATION OF FINES ...................................................................................... 484.5 SOURCES OF ERROR........................................................................................... 494.6 CONCLUDING DISCUSSION ................................................................................ 49

5 CONCLUSIONS ...................................................................................................... 51

6 FUTURE WORK .....................................................................................................53

7 REFERENCES.........................................................................................................55



12/124



13/124

1

1 INTRODUCTION

1.1 Blast Furnace Ironmaking

The blast furnace (BF) process is still the most common means of producing hotmetal. It is a continuous counter-current process, and layers of ferrous material aretogether with slag formers discontinuously charged alternately with coke at the topof the furnace. Slag formers such as BOF slag, limestone and quartzite form slagtogether with the gangue materials in the iron-bearing material and coke ash thatare insoluble in liquid iron. In the BF the coke has several functions. It acts as areducing agent, provides the energy required for endothermic reactions and formelting of iron and slag, and it has a mechanical function, in that it provides for

passage of liquids and gases in the furnace. Air is supplied in the blast to provide

the oxygen necessary to generate adequate amounts of CO(g) and H2(g) and tomaintain sufficient temperature for operation. In the water-cooled tuyeres, reducingagents such as pulverized coal are introduced together with the blast air. Coke and

pulverized coal, or other reducing agents, injected through the tuyeres arecombusted in the raceway, forming CO2. At temperatures above 1000C, CO2 isunstable in the presence of C and CO is generated according to the highlyendothermic Boudouard or solution loss reaction[1]

2COCCO2 + (1)

During the descent of material in the blast furnace shaft, the reduced iron picks upcarbon from the coke, which lowers the melting temperature. The temperature andCO content in the ascending reducing gas decrease on the way through the blastfurnace shaft. Because of the exchange of heat, the burden temperature increases asit moves down through the shaft. Coke remains solid at the tuyere level and in thehearth and provides a mechanical support in the bosh region, where the metal andslag are liquid. The liquid Femetis collected in the hearth. Hot metal and slag are

discontinuously tapped. Figure 1 shows a cross-section of a blast furnace.

1.1.1 Reactions in the blast furnaceThe blast furnace can be divided into different hypothetical zones.[1] The

boundaries between them are diffuse and the reduction reactions that dominate inone of the zones can occur at different positions in the blast furnace.



14/124

2

Figure 1Cross-section of a blast furnace

1.1.1.1 Upper zone

In the upper part of the blast furnace, the burden material temperature rises from

ambient to about 800C.

[1]

The temperature of the ascending gas decreases from800-1000C to between 100 and 250C. Hematite and magnetite are reduced tolower oxides. Indirect reduction of hematite by CO is exothermic.

24332 COO2FeCOO3Fe ++ (2)

H2also participates in the reduction of hematite

OHO2FeHO3Fe 243232 ++ (3)



15/124

3

CO or H2reduces magnetite by the endothermic reactions

243 CO3FeOCOOFe ++ (4)

OH3FeOHOFe 2243 ++ (5)

Carbon decomposition according to the reverse Boudouard reaction is mostpronounced at temperatures between 500 and 550C

CCO2CO 2 + (6)

1.1.1.2 Middle zoneIn the isothermal or thermal reserve zone, the temperatures of the solids and gas arein the range between 800 and 1000C.[1] In this zone, most of the indirectreduction, especially of wustite, occurs. The indirect reduction of wustite by CO isexothermic.

2met COFeCOFeO ++ (7)

The reduction step from wustite to Femetis by H2endothermic, but less endothermicat increased temperatures.

OHFeHFeO 2met2 ++ (8)

A chemically inactive zone, which is of major importance for a stable operation ofthe blast furnace, is found inside the isothermal zone, where the exchange ofoxygen between the ore and the gas is minor. Thus, the changes in gas compositionare also small.

Generation of H2takes place according to the water-gas shift reaction,

222 HCOOHCO ++ (9)



16/124

4

1.1.1.3 Lower zone

In the area from the tuyere level to a few meters above, called the bosh zone, themolten material reaches a temperature of 1400-1450C and the gas is cooled downto 800-1000C. C directly reduces oxides under the formation of CO. [1]Remainingwustite is directly reduced according to

COFeCFeO ++ (10)

The reaction is endothermic.

Combustion of reduction agents in front of the tuyeres results in an empty space inthe hearth periphery, which allows downward flow of the materials. The shape of

the combustion zone is of importance for a uniform gas distribution and burdendescent in the blast furnace.

1.1.1.4 Carbon deposition

At temperatures below 900C, carbon deposition can occur in Fe-O-C-H systemsdue to the decomposition of CO.[2] Addition of H2 will enhance the carbondeposition. CO content up to 20 percent does not influence the carbon deposition.Once the CO percentage increases beyond this value, the carbon deposition will berapid. Iron catalyses the decomposition of CO.[3]Presence of C or Fe3C is reported

to hinder the densification and sintering of iron phase.[4]Carburization of Femetorwustite grains can occur according to

CFeC3Fe 3+ (11)

23 COCFe2CO3Fe ++ (12)

23 4COCFe5CO3FeO ++ (13)

At temperatures above 900C, Fe3C reacts with wustite according to[5]

23 CO5FeCFe2FeO ++ (14)



17/124

5

1.1.2 Si content in hot metalSiO(g) is the dominant source of Si transfer in the blast furnace. SiO gas is formedin the high-temperature reaction of coke carbon on ash or slag silica.[6-7] To obtain alow silicon operation, the heat input should be lowered, to suppress the SiOformation. This is accomplished if the height of the dropping zone is lowered andraceway temperature is decreased. The carburized metal reacts with the ascendinggases containing SiO. As a result of increased PCR, the Si content in molten metaltends to increase, and it is assumed that the pulverized coal type will have aninfluence on the Si content.[8]

1.1.3 Pulverized coal injectionIn modern blast furnace ironmaking, continuous efforts are made to reduce coke

consumption by replacing coke with e.g., an increased amount of injectedpulverized coal. Injection of reducing agents is mainly done for economic reasons.Pulverized coal injection (PCI) was taken into operation in Sweden in 1985.Successful operation at pulverized coal injection rate (PCR) exceeding 200 kg/tHMis reported at blast furnaces.[9]

An increase in the PCR will among other things affect the composition of theascending reduction gases, the in-furnace temperature isotherms and possibly the

position of the cohesive zone. As the ore-to-coke ratio increases, so does the loadon the charged material. In-furnace isotherms for 1000C, determined with a feed-type vertical probe, showed higher temperature levels in the shaft with the increaseof PCI.[9]As a consequence, the cohesive zone will move upwards, resulting in anincrease in high-temperature furnace volume and a decrease in low-temperaturefurnace volume, and thus a possible decrease in the amount of indirect reduction of

pellets. Due to the amount of volatile matter in the injected coal, the H2content inthe reducing gas will increase as PCR increases.[10] Injection coals with differentamounts of volatile matters are used for blast furnace operation. A high-volatile

(HV) coal will generate more H2 compared to a low-volatile (LV) coal type.Investigations by Peters et al.[11]show that the amount of dust in the blast furnacegas increases with a higher PCR. The phenomenon is ascribable to higher gasvelocities caused by the increase in ore-to-coke ratio and the more fine-grainednature of burden compared with coke. At the same time, an increase in PCR willgenerate a higher amount of gas in the blast furnace.

According to Peters et al.,[11] the limits of PCR, from a process engineeringviewpoint, can mostly be expected as a consequence of incomplete combustion of



18/124

6

the coal during injection. Char is non-combusted leftovers of pulverized coal.Incomplete combustion will, however, not be limiting when char is consumed indirect reduction or carburization of iron. To ensure high combustibility at high

PCR, the design of the injection lances is an important consideration.

[9]

Aninjection method that creates a spatially uniform dispersion of the injected coalparticles promotes combustion efficiency, since it favours effective use of theoxygen for combustion near the coal particles. Yamaguchi et al.[12]concluded that afactor controlling the PCR is the discharge of unburnt char from the furnace top.

1.2 Conditions for Iron Oxide Reduction

Reduction of iron oxides can, as shown in section 1.1.1, both take place by CO andH

2. In the blast furnace, the CO content in the ascending gas is, due to the

Boudouard reaction, dependent on pressure and temperature. Decreasingtemperature and increasing pressure reduce the amount of CO in the gas inequilibrium with carbon. Reduction by H2 is independent of pressure. Figure 2shows the equilibrium conditions for reduction by CO and H2.

Figure 2The Fe-O-C (full) and Fe-O-H (dotted) equilibrium curves withthe superimposed Boudouard curve.[1]

In the blast furnace, reduction of iron oxides is simultaneously carried out with COand H2. H2 acts as a stronger reducing agent than CO at sufficiently high



19/124

7

temperatures relevant to the blast furnace process. Studies on reduction of sinteredore have shown that the reduction rates, especially in the temperature range

between 700C and 1000C, increased due to addition of H2.[13] Reduction of

commercial low silica pellets in a CO-CO2-N2 gas mixture over the temperaturerange 800-1100C has shown that the rate of reduction increases with a higherpartial pressure of CO.[14] The higher the temperature the faster was the rate ofreduction. Isothermal reduction of pellets in a bed with CO-H2atmosphere showedthat the overall reduction rates increased with increasing reduction temperature anddecreased with the degree of reduction.[15] At non-isothermal reduction of lowsilica hematite pellets in a CO-H2 atmosphere the reduction rate was faster at ahigher ratio of H2 in (CO+H2) mixture for a constant heating rate and constantinitial temperature.[16]For reduction in H2rich gas mixture, the reduction rate for

non-isothermal condition was lower than for isothermal reduction at correspondingtemperatures.

1.2.1 TexturesStudies by St. John et al.[17]have shown that the conditions for formation of porousiron depended on the gas composition, reaction temperature, oxide stoichiometryand the presence of impurity elements in solid solution with FeO. Reduction testson hematite pellets to form magnetite in a CO-CO2gas mixture were carried out by

Bradshaw et al.[18]

At 600C and pCO=0.25 atm a random distribution of pores wasobserved in the magnetite phase. An increase in the reaction temperature to 800Cand pCO=0.50 atm resulted in coarser pores and formation of lamella from thehematite/magnetite interface into the unreduced hematite. At 1000C and

pCO=0.125, no microporosity was apparent in the magnetite. All the investigatedpellets display an outer zone in which all but the largest grains were reduced tomagnetite. In the intermediate zone various stages of reduction were attained and inthe pellet core the hematite grains remained unattacked. Reduction of granularsolids of hematite in CO-CO2mixtures at a reaction temperature of 600C showed

random distribution of pores in the formed magnetite, a phenomenon not observedat 1000C.[19]Wustite formed at reduction of magnetite is dense, while wustite thatoriginates from hematite and oxidized magnetite is porous.[20]

During reduction of iron oxides, the initial reduction temperature is of majorimportance for the pore structure of the reduced iron.[21]The pore structure of Femet

becomes coarser with increasing temperature. A structure that is formed at a lowreduction temperature does not readily coarsen at subsequent temperature rise.Olsson et al.[22]showed that the pore structure of Fe

metformed during reduction of



20/124

8

iron oxides with H2 was a function of the reaction temperature. At a reductiontemperature of 900C the pore structure was homogenous. At 500C the porestructure was heterogeneous with large pores, which appeared to follow the initial

grain boundaries. Very fine secondary pores were observed through the individualgrains. Judging from the photomicrographs by Olsson et al. it can be concludedthat the average pore size in the Femet increases with increasing reductiontemperature. Wright[15]concluded, based on isothermal reduction of pellets in CO-H2 mixtures, that the structure of metallized pellets was highly temperaturedependent. At low reduction temperatures there was a great deal of internal

porosity in the individual Femetparticles and the interparticle pore sizes were verysmall. At increased reduction temperature, the voids progressively grow andcoalesce with the particle interstices and larger pores are formed.

In pellets of high original porosity there was a rapid reduction from hematite towustite without formation of distinct product layers.[21] The following reductionstep of wustite to iron was relatively rapid. The greater the driving force forreduction, the more pronounced was the formation of the product layers. In thecase of sintered hematite pellets the reduction proceeds topochemically during theformation of product layers.

1.2.2 Effect of impurity elementsLaboratory investigations by Geva et al.[23] have shown that the presence ofimpurity elements in solid solution in the iron oxide affects the final iron productmorphologies. Mg, Ti, Si, Ca, Na and K present together with FeO result indecreases in the CO concentrations necessary to obtain porous iron growth at anyreaction temperature relative to reactions on pure wustite. P has a marginal effecton the porous/dense iron transition and Al restricts the range of gas compositionsover which porous iron can be obtained. In H2/H2O mixtures, the presence of Ti,Mg, P, Si, Ca, K and Na favours porous iron formation. Additions of Al make the

formation of a porous iron layer difficult. The effects of impurities are additive. Naand K are present in the blast furnace and are generally undesirable, but can on theother hand play a significant role in improving the reducibility of the burden. Kfunctions as a catalyst for the reduction of iron oxide.[24]Since the solubility of K in

bulks of iron oxides has been reported not to exceed 0.1 wt%, much of the K isexpected to exist on the surface of iron oxides.[25]



21/124

9

1.3 Objectives

The objective of the performed studies was to develop an understanding of howdifferent blast furnace operation conditions will affect the reduction properties of

commercial pellets. Due to the increase in PCR, the conditions for pellet reductionchange as the ascending gas composition, the in-furnace temperature isotherms andpossibly the position of the cohesive zone are affected. Not only the PCR but alsothe volatile amounts in the injected coal and the method for oxygen supply to thelance may influence these properties in the blast furnace.

To learn more about pellet reduction under different blast furnace operatingconditions, laboratory reduction tests under blast furnace simulating conditionswere carried out for different PCR. Heating rate and temperature profiles were

based on mass and heat balance calculations and measurements in the LKABExperimental Blast Furnace (EBF). Conditions for formation of different pellettextures were determined. Pellet samples and process data from tests in the EBFduring operation at different PCR, injection of a LV and a HV coal and usingdifferent methods for oxygen supply were evaluated.

The hot metal Si content fluctuates in a blast furnace. Process data and pelletstaken out at the pellet charging at BF No. 3 at SSAB Tunnplt AB at Lule wereinvestigated in an attempt to find possible correlations between pellet reduction

properties and the hot metal Si.

Table 1 gives an overview of studies presented in Appendix I-III.



22/124

10

Table 1Overview of studies presented in Appendix I-III.

Laboratory Furnace EBF SSAB BF No. 3

Appendix

I

Reduction-BF simulating-Hypothetical PCR

Pellet texture

Effects of PCR andinjection coal type onconditions for pelletreduction

Properties of pelletsreduced duringdifferent PCR andinjection coal types

Raw materialsampling

Pellet properties

Appendix

II

Reduction-Simulated PCRbased on measure-ments in the EBF

Pellet textures

Process data

A

ppendix

III

Reduction-Simulated PCRbased on measure-ments in the EBF

Pellet textures

Evaluation of pellettextures fromoperation duringdifferent PCR,injection coal typesand methods foroxygen supply.



23/124

11

2 MATERIALS AND METHODS

2.1 Materials

Commercial olivine pellets were used in the investigations performed. Materialsamplings for the laboratory investigations were carried out at the LKAB

pelletizing plant at Malmberget and at the SSAB BF No. 3 pellet-charging stream.The chemical composition of the olivine pellets MPBO and KPBO produced byLKAB can be seen in Table 2. MPBO was used in the EBF trial.

Table 2Chemical composition of MPBO and KPBO in percent[26]

Pellet Fe FeO CaO SiO2 MgO Al2O3 CaO/SiO2MPBO 66.8 0.5 0.35 1.7 1.5 0.32 0.21

KPBO 66.6 0.4 0.46 2.0 1.5 0.23 0.23

In the EBF tests two different coal types were used: a LV containing 19.6 percentvolatile matters and a HV containing 38.0 percent volatile matters. Other additionscorresponded to operation at the SSAB BF No. 3, except for quartzite, which wascharged to provide for the SiO2in the slag and reach the desired slag volume.

2.2 SSAB Blast Furnace No. 3

SSAB Tunnplt AB has had blast furnaces operating at the Lule plant since 1951.In August 2000 the new BF No. 3, which replaced BF No. 1 and BF No. 2, wastaken into operation with a hot metal production of 2.2 Mt/year. Design data for theSSAB BF No. 3 are listed in Table 3. The ferrous burden consists entirely ofolivine pellets from Malmberget, MPBO, and Kiruna, KPBO, produced by theSwedish iron ore producer LKAB.

2.2.1 Raw material sampling

Raw material samplings were carried out at the pellet feeders close to the blastfurnace. Pellets were taken out every second hour. Process data, including the hotmetal Si content, were stored at the same time. Low temperature reductiondisintegration tests were performed and the reduction strength determined, see

paragraph 2.5.1.



24/124

12

Table 3Design data, Blast Furnace No. 3 at SSAB Tunnplt AB, Lule[27]

BF No. 3Year of erection 2000 YearDesign SSAB/Kvaerner

Start of campaign Aug. 25th

2000Hearth diameter 11.4 mWorking volume 2400 m3

Total volume 2850 m3

No of tuyeres 32No of tap holes 2Top pressure 150 kPa

Daily output 6700 tonnes

Charging equipment Belt/Bell less top Central

feed

2.3 LKAB Experimental Blast Furnace

The EBF at MEFOS has a working volume of 8.2 m3, a diameter at tuyere level of1.2 m and is equipped with a system for injection of reduction agents.[28]The heightfrom tuyere level to stock line is 6 m and there are three tuyeres separated by 120.After a campaign, the furnace can be N2quenched and excavated. The process isinterrupted, nitrogen throughput from the top started, followed by a decreased and

finally stopped blast volume. Dissection of the EBF can start after at least ten daysof cooling and is carried out like an archaeological excavation, where basket-samples introduced in the final hours of operation are recovered. A schematicdrawing of the EBF plant including raw material handling, injection system andgas cleaning system can be seen in Figure 3.

During operation, in-burden probes were used for sampling of the burden and forthe measurement of the horizontal temperature- and gas profiles. Figure 4 show the

positions of the shaft probes in the EBF. The shaft probe material was divided in

sub-samples. During ideal conditions, a packed probe will generate 5 sub-samplesfor the upper probe and 6 for the lower. Usually, only 3-4 sub-samples weregenerated for each probe and it was assumed that the sub-sample with the highestnumber was taken close to the wall.

Estimation of the positions of the in-furnace temperature isotherms was done byallowing a thermocouple to descend with the burden.



25/124

13

Figure 3Drawing of the EBF plant

Figure 4Drawing of the EBF with shaft probes indicated



26/124

14

2.3.1 EBF tests2.3.1.1 Influence of pre-set process conditions

The influence of PCR, coal type and method for oxygen supply on olivine pelletreduction properties was investigated during an EBF test. The PCI system was anoxy-coal system and the oxygen added to the lance could be replaced by air ifdesired. In the tables and figures, operation with oxygen added to the lance istermed Oxy Coal and operation with air addition to the lance termed Air Coal.Horizontal gas composition and temperature profiles were measured at the

positions of the shaft probes. Pre-set process conditions during the 6 test periodsare presented in Table 4.

Table 4 Overview of the pre-set process conditions during the EBF trial

Coal type PCR (kg/tHM) Method for oxygen supply1 LV 152 Oxy Coal2 LV 146 Air Coal3 LV 79 Oxy Coal4 HV 94 Oxy Coal5 HV 152 Oxy Coal6 HV 143 Air Coal

Process parameters

Test Period

In-burden material was taken out with the shaft probes and the pellet properties

were evaluated. During the test periods, process data including burden decent rateswere stored. Pellet strength and fines generation were studied based on samplesfrom EBF operation during oxygen addition to the coal lance.

2.3.1.2 Basket samples

Basket samples were introduced into the EBF prior to quenching. About 600 gramsof pellet were put into a basket constructed of a high-temperature resistant metallicnet. Pellets for the basket samples were taken out during the raw material sampling

period at SSAB Tunnplt AB. For further details on the pellet selection, see section3.1.

2.4 Laboratory Reduction Furnace

2.4.1 Finishing of laboratory reduction furnacePrior to the laboratory reduction experiments, considerable improvements of theformer reduction furnace system were made. The furnace at Lule University ofTechnology (LTU) is a vertical steel tube-type furnace with an inner diameter of

about 6 cm that is heated electrically by U-shaped Super-Kanthal elements with a



27/124

15

heating zone of about 8 cm in height. The experimental apparatus is shown inFigure 5. Hardware and software for simultaneous temperature and gas controlwere developed and put into operation. The design of the Furnace Control program

enables use of blast furnace simulating conditions with optional choice of heatingrate and gas composition as well as isothermal test programs using a constant orvaried reduction gas composition. Test runs were made for adjustment of the PIDcontroller parameters and measurement of the temperature profile inside thevertical steel tube. During the reduction tests, the pellet samples were placed in the10 cm zone in which the temperature gradient was 2-3C. The gas supply system isequipped with digital Multi-Bus Flow-Bus regulators for input gas flows of CO,H2, CO2and N2. The gas is introduced in the bottom of the tube and heated in a bedof Al2O3balls. A thermocouple for temperature measurement is introduced from

the bottom of the tube and situated approximately 20 mm below the sample, whichis suspended in the balance with metal wires. Test runs were made to investigatethe influence of input flow on pellet reduction in blast furnace simulating tests.Input gas composition and temperature are controlled by a computer, which also isused to store data with a frequency chosen for the parameters.

Figure 5 Schematic view of the experimental apparatus used for laboratory reduction tests

2.4.2 Experimental procedureAt the test starting temperature, a nitrogen flow at 12 l/min was introduced. Thetemperature and N2gas flow were held at constant values for a few minutes beforethe sample was introduced into the furnace. Each sample consisted of dry pelletswith a starting weight of 70 to 85 grams. The sample material was placed in a



28/124

16

basket. As soon as the sample had been introduced into the constant temperaturezone, the test was started and the temperature program started at the same time asthe gas composition was changed into a reducing atmosphere. After the test, which

could be interrupted at any optional point the sample was cooled in pure N2in thewater-cooled top. Total gas flow was maintained at 12 l/min during the entire test.The reduction degrees attained in the laboratory reduction tests were calculatedaccording to

100(g)oxygenoriginal

(g)removedoxygen(%)degreereduction = (15)

It was assumed that the total weight change during reduction was caused byremoval of oxygen.

2.4.3 Laboratory reduction conditions2.4.3.1 Simulated BF

One blast furnace simulating reduction profile was based on a normal test programused at LKAB.[29]Heating rate and reduction gas composition for this program are

presented in Figure 6.

500

600

700

800

900

1000

1100

1200

0 30 60 90 120 150 180 210 240

Time (min)

Temperature(C)

0

10

20

30

40

%Temperature

CO

CO2

H2

Figure 6Blast furnace simulating reduction profile based on a normaltest program at LKAB[29]

Pellets taken out during the material sampling period at SSAB Tunnplt AB werereduced in the simulated BF reduction profile. The tests were interrupted at a



29/124

17

temperature of 1025C. An overview of the selection of pellets for laboratoryreduction from the sampling occasions is given in section 3.1.

2.4.3.2 Hypothetical PCR

Two different temperature profiles were used in the study of the influence ofhypothetical PCR on pellet reduction properties. The slow heating rate was basedon a normal test procedure at LKAB[29]and the fast heating rate was based on avertical temperature measurement made in the EBF during a PCR of approximately130 kg/tHM. The gas compositions used for the different hypothetical processconditions with all coke operation and hypothetical PCR were calculated usingmass and heat balances for estimation of the gas composition at various levels inthe blast furnace. The test conditions for hypothetical All Coke and a hypothetical

PCR of 200 kg/tHM are shown in Figure 7 and Figure 8.

500

600

700

800

900

1000

1100

1200

0 20 40 60 80 100 120

Time (min)

Temperat

ure(C)

0

10

20

30

40

50

60

%

Temperature

H2, 200 kg/tHm

H2, All Coke

CO, 200 kg/tHMCO, All Coke

CO2, 200 kg/tHM

CO2, All Coke

Figure 7Reduction profiles for calculated hypothetical All Coke and a PCR at 200kg/tHM. Fast heating rate based on a vertical temperature measurement in the EBF



30/124

18

500

600

700

800

900

1000

1100

1200

0 20 40 60 80 100 120 140 160 180 200 220 240

Time (min)

Temperature(C)

0

10

20

30

40

50

60

%

Temperature

H2, 200 kg/tHm

H2, All Coke

CO, 200 kg/tHm

CO, All coke

CO2, 200 kg/tHm

CO2, All coke

Figure 8Reduction profiles for calculated hypothetical All Coke and a PCR at 200kg/tHM. Slow heating rate based on a normal test procedure at LKAB

Table 5 shows an overview of laboratory reduction experiments with hypotheticalPCR. Olivine pellets produced at Malmberget were used in tests 1-6.

Table 5Schematic overview of blast-furnace-simulating reduction experiments performedwith hypothetical PCR

Slow Fast All-Coke PCR 146 kg/tHM PCR 200 kg/tHM1 X X2 X X3 X X4 X X5 X X6 X X

TestTemperature Profile Gas Profile



31/124

19

2.4.3.3 Simulated PCR

Heating rate and gas profiles for the simulated PCR profiles were estimated frommeasurements made in the EBF during operation at the high and low PCR in test

periods 1 and 3. Oxygen supply was by oxy coal and the LV coal type wasinjected. The heating rate profiles were estimated from:

Vertical temperature measurements Average burden decent rates Horizontal temperature profiles at the position of the shaft probes CO/CO2 ratio at the position of the lower shaft probe for temperature

estimation from an oxygen potential diagram

For the high and the low PCR, a fast heating rate was estimated to simulate a centreprofile and a slow heating rate was estimated to simulate an intermediate/wallprofile. The reduction gas compositions were estimated from the top gas analysisand the gas composition at the position of the upper and lower shaft probes. Thegas compositions used together with the fast heating rates were estimated from thecentre gas composition in the EBF. For the slow heating rates, the gascompositions at an intermediate position in the EBF were the basis of the reductiongas composition. N2was filled up to a gas flow of 100 percent. Test conditions areshown in Figure 9 and Figure 10.

500

600

700

800

900

1000

1100

1200

0 20 40 60 80 100 120 140

Time (min)

Temperature(C)

0

10

20

30

40

50

%

Fast, Temperature

Slow, Temperature

Fast, CO

Fast, CO2

Fast, H2

Slow, CO

Slow, CO2

Slow, H2

CO

H2

CO2

Figure 9 Heating rate and gas composition profiles for simulated high PCR.Heating rate and gas profiles estimated from measurements in the EBF



32/124

20

500

600

700

800

900

1000

1100

1200

0 20 40 60 80 100 120 140

Time (min)

Temperature(C

)

0

10

20

30

40

50

%

Fast, TemperatureSlow, Temperature

Fast, CO

Fast, CO2

Fast, H2

Slow, CO

Slow, CO2

Slow, H2

CO

H2

CO2

Figure 10Heating rate and gas composition profiles for simulated low PCR.Heating rate and gas profiles estimated from measurements in the EBF

One set of experiments was interrupted at a reduction degree of approximately 40percent and one set was interrupted at a furnace temperature of 1100C. Table 6gives an overview of the simulated PCR experiments performed. Olivine pellets

produced at Malmberget were used in tests 7-14.

Table 6Schematic overview of BF PCR simulating experiments performed

High PCR Low PCR High PCR Low PCR7 X X X8 X X X9 X X X10 X X X11 X X X12 X X X

13 X X X14 X X X

SlowPCR

TestTest End

High LowRed. Degree

40 %Temp.1100C

Heating RateFast



33/124

21

2.5 Evaluation of Pellet Properties

2.5.1 Pellet strengthLow-temperature reduction disintegration (LTB) and the high-temperature

breakdown (ITH) tests were carried out on a selection of pellets taken out duringthe raw material sampling period at SSAB Tunnplt AB. LTB was performed byLKAB according to ISO13930[26], except for the pellet size fraction, where pelletsfrom the total samples were used instead of the 10-12.5 mm fraction. ITH was

performed by LKAB and determined by tumbling in a drum with a length of 700mm and an inner diameter of 130 mm at 20 rpm for 600 turns after reductionaccording to ISO 4695[30]. Pellets from the total samples were also used in thiscase. After tumbling, the sample was sieved for +6.3 mm and 0.5 mm and the ITH

result was presented as the percent fraction +6.3 mm and 0.5 mm.

The strength of pellets taken out with the upper shaft probe in the EBF was testedin an I-drum with a diameter of 130 mm and a length of 700 mm. 300 grams of

pellets in the size range 10-12.5 mm were rotated 600 times at 20 rpm.

2.5.2 Pellet texturesPellet textures were investigated by light optical microscopy (LOM) and byscanning electron microscopy (SEM). Prior to observation, pellets were coldmounted in epoxy resin. In order to observe the largest pellet cross section, themounts were cut at the maximum pellet diameters. The surface of the section was

polished. A Nikon E600 POL polarizing microscope was used for LOM. For SEM,a Philips XL 30 equipped with Energy Dispersive X-ray Analysis (EDS) forchemical mapping was used. Prior to SEM investigation, the cross sections of themounted pellet samples were coated with a thin layer of gold using a Bal-Tec MCS010 sputter coater.

2.5.3 Identification of trace elements in pelletsIn order to identify trace elements in the pellet periphery and core, pellets wererotated in an I-drum at 600 revolutions for 30 minutes. After I-drum treatment,

pellet fragments in the size range 212m 1,5 mm and >10 mm were separatelyground in a ring mill and X-Ray diffraction was carried out on the groundmaterials. A Siemens X-ray diffractometer was used to record scattering intensitiesof samples by using a Copper K radiation (40 kV, 40mA) as the X-ray source.The samples were continuously scanned over an angular range of sin210-90 by



34/124

22

using a step size of 0.020 and over sin235-55 with a step size of 0.020 and astep time of 6 seconds at each step.

Identification of trace elements in pellets was performed on sub-samples 3 takenout with the lower shaft probe during the EBF test. Investigations on the pelletcore, >10 mm fragments, were carried out for the low PCR periods.

2.6 Chemical Analysis

Chemical analyses were provided by SSAB Tunnplt AB and LKAB. X-rayfluorescence and LECO combustion analysis were used at SSAB Tunnplt AB andwet chemical titration at LKAB.



35/124

23

3 RESULTS

3.1 Evaluation of Process Data and Pellet Samples from SSAB

Blast Furnace No. 3, Appendix I3.1.1 Process data and selection of pellets for further testingDuring the raw material sampling period, the Si content in the hot metal reached

between 0.15 and 0.54 percent and legible decreases were observed betweenobservations 19 to 28 and 38 to 49, see Figure 11. The 74 Si observationscorresponded to measurements during 48 hours. Samples of MPBO and KPBO forfurther testing were chosen during a period of increase and decrease of the Sicontent in hot metal. Sampling occasions A-H and the relation to the hot metal Sicontent are marked in Figure 11. Figure 12 shows the PCR in relation to the hotmetal Si during the period in question. The time delay between pellet charging andhot metal tapping was estimated to approximately 8 hours. Pellets charged inconnection to pellet sampling occasion A influenced the hot metal Si content 8hours later.

LTB and ITH tests were performed on MPBO and KPBO samples A-H. Sampleswith the lowest share, highest share, and median share of the >6.3 mm fractionwere chosen for further investigation. MPBO samples B, E and G, respectively,

and KPBO samples B, F, and G, respectively, were chosen.

0

0,1

0,2

0,3

0,4

0,5

0,6

0 10 20 30 40 50 60 70 80

Observation

%Si

A CB ED F G H

Figure 11Si content in hot metal during the raw material samplingperiod and A-H samples used for metallurgical testing



36/124

24

0

0,1

0,2

0,3

0,4

0,5

0,6

37353,8 37354,3 37354,8 37355,3 37355,8 37356,3 37356,8

%Si

100

110

120

130

140

150

160

170

180

190

200

210

220

PCR(kg/tHM

)

Si PCR

Figure 12PCR and the relation to hot metal Si content duringthe raw material sampling period

3.1.2 Laboratory reduction under simulated BF conditionsReduction profiles of the MPBO samples B, E and G and KPBO samples B, F andG indicated almost the same reduction behaviour, see Figure 13.

0

10

20

30

40

50

60

0 30 60 90 120 150 180 210 240

Time (min)

ReductionDegree(%)

MPBO B

MPBO E

MPBO G

KPBO B

KPBO F

KPBO G

Figure 13 Reduction profiles for laboratory reduced MPBOsamples B, E, G and KPBO B, F and G

3.1.2.1 Pellet textures attained

After reduction, Femetdominated in the pellet peripheries, while a few small areaswere found in the pellets core. Wustite dominated the cores and some magnetitewas observed. Differences in Femettexture were observed between the pellet types.Compared to the KPBO, the MPBO samples showed larger areas of coherent Femet



37/124

25

in the pellet peripheries samples. Figure 14 shows typical textures of the MPBOand KPBO pellets.

Core Periphery

MPBOE

KPBOF

Figure 14 Typical pellet textures of MPBO (upper) and KPBO (lower) observed inLOM after blast furnace reduction simulation tests. Pellet core (left) and periphery(right). Magnetite = Smooth grey, Wustite = Broken grey, Femet = White, Pores =Black

3.1.3 Basket samplesFrom an excavation of the EBF, basket samples containing material from the samesamples as MPBO sampling B, E and G and KPBO sampling B, F and G wererecovered directly below the position of the upper and lower shaft probes. All of

the basket samples containing corresponding material as in the simulated blastfurnace tests showed similar iron oxide and Femet textures after reduction in theEBF. The reduction degrees attained were higher in basket samples recoveredfurther down in the EBF. The size and number of Femet areas increased withincreasing reduction degree.



38/124

26

3.2 Laboratory Reduction for Hypothetical PCR, Appendix I

Figure 15 shows that an increase in the hypothetical PCR increased the degree ofreduction at any temperature once the reduction degree had reached about 10

percent; a pattern that was independent of temperature profile. The most apparentdifference in reduction behaviour was observed between a hypothetical All Cokecase, tests 1 and 4, and a hypothetical PCI at 146 kg/tHM or 200 kg/tHM, tests 2 or3 and 5 or 6. A clear difference between the hypothetical PCR of 146 and 200kg/tHM, tests 2 and 3, was observed for the slow temperature profile attemperatures above 750C.

0

10

20

30

40

50

60

70

0 30 60 90 120 150 180 210

Time (min)

ReductionDegree(%)

500

600

700

800

900

1000

1100

Temperature(C)

1 All Coke

2 146 kg/tHM

3 200 kg/tHM

4 All Coke

5 146 kg/tHM

6 200 kg/tHM

Temperature,slow

Temperature,fast

Temperature, slowTemperature, fast

Figure 15Temperature- and reduction profiles for laboratory reductiontests simulating hypothetical All Coke and different PCR

Typical textures from the study of pellets reduced in laboratory tests based onhypothetical All Coke and hypothetical PCR tests are presented in Figure 16.Significant differences in the iron oxide and Femettextures were not observed when

pellets from tests with hypothetical PCR of 146 and 200 kg/tHM and

corresponding temperature profiles were compared. An increase of the hypotheticalPCR for the slow temperature profile increased the amount of Femet in the pelletcore. Wustite dominated in the pellet cores with few Femet areas. For thehypothetical All Coke, magnetite dominated the pellet core and wustite and Femetthe periphery. Magnetite was observed through the entire cross section of pellets atthe same time as Femetand wustite dominated in the pellet periphery for pellets athypothetical PCR and a fast temperature profile. For the hypothetical All Coke,magnetite dominated in the core. Femet was observed together with wustite andsome magnetite in the pellet periphery.



39/124

27

Core Periphery

1

3

4

6

Figure 16 Pellet textures observed in LOM after hypothetical All Coke (1, slowtemperature profile and 4, fast temperature profile) and PCR of 200kg/tHM (3, slowtemperature profile and 6, fast temperature profile). Pellet core (left) and periphery(right). Magnetite = Smooth grey, Wustite = Broken grey, Femet= White, Pores = Black



40/124

28

3.3 EBF Tests, Appendix I-III

3.3.1 Ascending gas compositions, Appendix I and IIIThe H2-content in the reduction gases was observed to increase with increasing

PCR. Oxygen supply to the coal lance as well as operation using the HV coalresulted in a higher H2amount as compared to when air was added to the lance andwith operation using the LV coal. The H2content increased during operation withthe HV coal, except during test period 4, when lower temperatures were observedin the EBF. The H2content was at a generally higher level at the lower shaft probein the EBF, due to the higher temperature.

TheCO[1]level,

)%CO/(%CO100%CO% 22CO += , (16)

was observed to reach a lower level at the position of the upper shaft probe duringoperation using the LV coal compared to operation using the HV coal. During thetests, the reduction potential of the gas was higher in the centre of the furnace anddecreased towards the periphery.

3.3.2 Influence of coal type and PCR, Appendix IThe particle size distribution of material taken out with the upper shaft probeduring high and low PCR of the HV and LV coal type can be seen in Figure 17.Figure 18 shows the content of Fe and C in the


41/124

29

0

2

4

6

8

10

12

14

16

%6.3mm

0-0,5 mm 0,5-3,3 mm 3,3-6,3 mm >6.3 mm

LV Coal HV Coal

High PCR Low PCR High PCR

Figure 17Particle size distribution in the total material samplestaken out with the upper shaft probe in the EBF for high and lowPCR during injection of the HV and LV coal

0

10

20

30

40

50

60

15055 15056 15062 15070 15072 15080 15082

%in6.3 mm were found at an increased reduction degree. This resultindicated low disintegration, but at the same time the highest amount of abrasion inthe 6.3 mm


44/124

32

3.3.3.2 Pellet textures

Magnetite dominated the cores of pellets taken out with the upper shaft probeduring injection of the LV coal in test periods 1-3. Some hematite and wustite werealso observed in the pellet cores. The pellet peripheries were mainly made up ofwustite. Wustite and magnetite dominated in the pellet cores in pellets taken outwith the upper shaft probe during injection of the HV coal in test periods 4-6. Afew areas of Femetobserved in the pellet core increased in size and number towardsthe periphery. Wustite and Femetdominated the periphery. The transition betweenthe texture in the pellet core and periphery was blurry and any distinct productlayers were not formed. Typical textures of pellet cores and peripheries obtained in

pellets taken out with the upper shaft probe during test periods 1-6 are shown inFigure 23 and Figure 24.

In pellets taken out with the lower shaft probe, Femet dominated the texture.Injection of the HV coal type increased the pore size in the periphery. Comparingthe pellet cores, the size of the continuous Femetareas increased during injection ofthe HV coal compared to injection of the LV coal at high PCR and independent ofoxygen supply method. In the pellet core from period 1, softened wustitedominated. Typical textures of pellet cores and peripheries obtained in pellets takenout with the lower shaft probe are shown in Figure 25and Figure 26.

EDS mapping of pellet core, intermediate area and pellet periphery taken outduring operation with the LV and HV coal type, test periods 2 and 5, are presentedin Figure 27 and Figure 28. Slag areas were found next to or enclosed in the Femetin the entire cross section in pellets taken out with the lower shaft probe during test

periods 1-6. The size and distribution of the slag areas and areas with increased Kcontent varied between the test periods. The tendency for liquid formation, as can

be seen when Figure 27 and Figure 28 are compared, was more pronounced inpellets taken out during injection of the HV coal type, as can be seen by the smooth

slag areas compared to the diffuse slag areas during LV coal operation. Sometimes,the pellet periphery was covered by a layer consisting of mainly Mg, Si, K, Ca andAl.

K was observed in pellets taken out with the lower shaft probe, independent of pre-set process conditions. Areas of increased K content were in the core of the pelletsto be found next to or surrounded by Femet. In pellets taken out during injection ofthe LV coal type, K was located next to Femet, while Femet surrounds K afterinjection of the HV coal type. In the pellet periphery, K was found between Femetas



45/124

33

well as in the surface cover of pellets, independent of the pre-set process conditionsduring sampling. K formed compounds with Si, Al, Mg, and in a few cases K andTi coexist. The coexistence of K and the slag elements was more frequent in the

pellet periphery, and especially in the surface cover, compared to the core of thepellets. Ca-Si slag was observed in the pellet cores and coexistence between theelements was observed in the pellet periphery as well as in the surface cover. Alwas mainly observed in the surface cover during injection of the LV coal type,though throughout the entire pellets during HV coal injection. In the surface coverthe most prominent areas of coexistence of K, Si, Mg and Ca were observed.



46/124

34

Core Periphery

1

2

3

Figure 23Typical textures, observed in LOM, of pellets taken out with the upper shaftprobe in the EBF during test period 1-3. Pellet core (left) and pellet periphery (right).Hematite = Light grey, Magnetite = Smooth grey, Wustite = Broken grey, Pores = Black



47/124

35

Core Periphery

4

5

6

Figure 24Typical textures, observed in LOM, of pellets taken out with the upper shaftprobe in the EBF during test period 4-6. Pellet core (left) and pellet periphery (right).Hematite = Light grey, Magnetite = Smooth grey, Wustite = Broken grey, Femet= White,

Pores = Black



48/124

36

Core Periphery

1

2

3

Figure 25Typical textures, observed in LOM, of pellets taken out with the lower shaftprobe in the EBF during test period 1-3. Pellet core (left) and pellet periphery (right).Wustite = Grey, Femet= White (lightest), Pores = Black



49/124

37

Core Periphery

4

5

6

Figure 26 Typical textures, observed in LOM, of pellets taken out with the lower shaftprobe in the EBF during test period 4-6. Pellet core (left) and pellet periphery (right).Wustite = Grey, Femet= White (lightest), Pores = Black



50/124

38

Core Intermediate Periphery

Fe

Mg

K

Ca

Si

Al

Figure 27 EDS mapping of core, intermediate area and periphery ofpellet taken out with the lower shaft probe during operation with the LVcoal type in test period 2



51/124

39

Core Intermediate Periphery

Fe

Mg

K

Ca

Si

Al

Figure 28 EDS mapping of core, intermediate area and periphery ofpellet taken out with the lower shaft probe during operation with the HVcoal type in test period 5



52/124

40

3.3.3.3 Pellet fragments

Chemical analysis of the pellet fragments in the size range 212m 1.5 mmshowed an increased carbon content after operation with injected HV coal.Increased contents of K and Ca, see Figure 29, are also observed. In the XRD-analysis Fe3C was observed in samples from test periods 2-6.

0

1

2

3

4

5

6

7

8

9

10

Weight-%

0

10

20

30

40

50

60

70

80

90

100

Fe%

C K Si Ca

Na Al Fe

High PCR Low PCR High PCR

Oxy Coal Air Coal Oxy Coal Air Coal

LV Coal HV Coal

Figure 29Element distributions in the fraction 212m 1.5 mmafter I-drum treatment of pellets from sub-sample 3 taken out withthe lower shaft probe during operation in the EBF

For the low PCR, the chemical analysis of the >10 mm pellet fragments showed anincreased amount of carbon after operation with the injected HV coal type. In thediffractogram, Fe3C was observed after operation with the injected HV coal typeand FeO was observed after injection of the LV coal type. The pellet fragmentswere mainly made up of Femet.

3.3.4 Laboratory reduction with simulated PCR, Appendix II and IIIFigure 30 shows the reduction profiles for tests 7-14. For the same level of

simulated PCR, at the test end temperature of 1100C, a higher reduction degreewas attained for the slow heating rates when tests 10 and 8 and tests 14 and 12,were compared. After reduction simulating the fast heating rate profiles, tests 8 and12, approximately equal reduction degrees were attained independent of simulatedPCR. A higher reduction degree was attained for the slow heating rates at the endof test 10, simulated high PCR, compared to test 14, simulated low PCR. In tests 7,9, 11 and 13, the reduction was interrupted at a reduction degree of approximately40 percent.



53/124

41

0

10

20

30

40

50

60

70

80

90

0 20 40 60 80 100 120 140

Time (min)

ReductionDegree

(%)

7, High PCI, Fast

8, High PCI, Fast

9, High PCI, Slow

10, High PCI, Slow

11, Low PCI, Fast

12, Low PCI, Fast

13, Low PCI, Slow

14, Low PCI, Slow

8

10

12

14

Figure 30Reduction profiles for laboratory tests 7-14. Test 8, 10,12 and 14 interrupted at the test end temperature of 1100C andtests 7, 9, 11 and 13 at an attained reduction degree ofapproximately 40 percent

3.3.4.1 Pellet textures

Significant differences in the iron oxide textures are observed in the pellet corewhen pellets from the reduction tests simulating the high PCR (see Figure 31) andlow PCR (see Figure 32) are compared. A grain texture was observed in the pelletcore of samples after tests 11-14, simulating the low PCR, an observation that was

not discernible after tests 7-10, simulating the high PCR. Textures from the pelletcore and periphery from tests 7-14 are presented in Figure 31 and Figure 32.Together with the areas of Femetin the pellet periphery, a few areas of iron oxidewere present after tests 8 and 10, which were full-length tests simulating the highPCR. After full-length tests simulating the low PCR, tests 12 and 14, thecorresponding areas in the pellet periphery showed noticeable areas of iron oxidesurrounded by Femetwithin the texture dominated by Femet. A few areas of Femetwere observed in the pellet core from test samples interrupted at the test endtemperature of 1100C. Femetwas not observed in pellets reduced to a reductiondegree of approximately 40 percent. Significant differences in the Femettexture ofthe pellet periphery, independent of final reduction degree, were not observed

between pellets reduced according to similar reduction profile. The distributionbetween the pellet core, mainly made up of iron oxide, and the pellet peripherydominated by Femet varied, however, in accordance with the reduction degreesattained.



54/124

42

Core Periphery

7

8

9

10

Figure 31Pellet textures observed in LOM after reduction tests 7-10. Pellet core (left)and pellet periphery (right). Iron oxides = Grey, Femet= White, Pores = Black



55/124

43

Core Periphery

11

12

13

14

Figure 32Pellet textures observed in LOM after reduction tests 11-14. Pellet core(left) and pellet periphery (right). Iron oxides = Grey, Femet= White, Pores = Black



56/124

44



57/124

45

4 DISCUSSION

4.1 Reduction

Laboratory reduction experiments at LTU were carried out in three different ways.Conditions for blast furnace simulating tests were based on a normal test programat LKAB.[29] For the hypothetical PCR tests the reduction gas composition atvarious temperatures was calculated using heat and mass balances. The test starttemperature was independent of temperature profile and hypothetical PCR. Thesimulated PCR tests were based on measurements in the EBF during differentPCR. The test start temperature was not constant for all of the simulated PCR tests.

An increase in PCR, resulting in an increased reduction potential and increase inthe temperature level, compensated for the loss in reduction time between the slowand the fast heating rates. Generally, a higher PCR resulted in a higher pelletreduction degree at an equal reduction time. However, the initial reductiontemperature and heating rate were of importance for the reduction degrees attained.A higher start reduction temperature, as in the simulated high PCR, resulted in afaster initial reduction compared to a lower start temperature. On the other hand, afaster heating rate compensated for a lower start temperature and lower simulatedPCR. Although there were differences in PCR, approximately equal reduction

degrees were attained for the simulated low PCR and fast heating rate andsimulated high PCR and slow heating rate at a reduction time exceeding 35minutes. In general, an increased reduction time increases the final reductiondegree at the end of the test. A small addition of H2to the reduction gas, comparedto reduction without H2, had a more prominent effect on the pellet reduction thanan increase in the H2content.

In the EBF, pellets in samples taken out with the upper shaft probe have, as wasconfirmed by the chemical analysis as well as by LOM, during injection of the HV

coal reached a higher reduction degree compared to during operation with the LVcoal. In pellets taken out with the lower shaft probe significant differences inaverage reduction degree were not found. The results indicated that the testconditions in terms of the reduction potential of the gas, temperature profile andtime for all cases resulted in a thermal reserve zone that was sufficient forreduction of pellets under all the investigated process conditions.

Pellet samples, which showed differences in LTB and ITH, indicated almost the

same reduction behaviour. In the simulated blast furnace laboratory tests the



58/124

46

conditions were equal and differences in LTB and ITH obviously did not correlateto the reduction behaviour.

4.2 TextureThe initial reduction conditions, in terms of temperature and gas composition, wereconclusive for the pellet texture formed. In the simulated PCR tests, differences inthe pellet textures that were observed in samples after reduction to 1100C werealready found at a reduction degree of 40 percent. The thickness of the peripheralzone, in which Femet dominates, increased with increasing reduction degree butdistinct boundaries between the pellet core and periphery were not observed. Thetests with simulated high PCR and slow heating rate and simulated low PCR andfast heating rate showed, despite similar reduction behaviour above a reductiondegree of 25 percent, differences in pellet texture. These results support the claimthat the texture differences were most likely to occur in the beginning of thereduction. A simulated low PCR resulted in a grain texture in the pellet core thatwas not observed at simulated high PCR. In the pellet periphery, iron oxides weresurrounded by Femetand the Femettexture was denser at low PCR. The size of the

pores in the pellet periphery was larger at high PCR. A higher start reductiontemperature and a more rapid temperature increase in combination with an increasein the reduction potential of the gas counteracted the formation of a grain texture in

the pellet core. At the same time, a porous Femettexture was formed in the pelletperiphery.

The present results indicate that the initial reduction temperature is more importantfor the pellet texture formed than the gas composition. Significant texturedifferences, which most likely were absent due to identical test start temperatures,were not observed in the pellet core after hypothetical PCR reduction in thelaboratory furnace. The pellet peripheries showed a similar texture, except for thehypothetical PCR of 200 kg/tHM and slow temperature profile, where the areas ofFemetbecame denser. No distinct boundaries between the pellet core and peripherywere observed in the pellets from the hypothetical PCR laboratory experiments.

In pellets taken out with the upper shaft probe during the EBF tests, Femetwas notobserved during injection of the LV coal. During HV coal operation, a few areas ofFemetwere observed in the pellet core, increasing in size and number towards the

periphery. The results from pellets taken out with the lower shaft probe during theEBF tests indicated that a combination of a high temperature level and high H2



59/124

47

content in the reduction gas were two important parameters for the formation ofsmooth Femet areas. A straggling Femet texture was formed at lower reductiontemperatures and H2 levels in the reduction gas. The pores of the Femet textures

throughout the pellets and the solid Femetareas in the pellet cores during injectionof the HV coal were larger compared to operation with LV coal. The marked Femetareas were observed to be more compact during operation with the HV coal typecompared to the LV coal type. The straggling Femettexture was, especially in the

pellet periphery, identified during operation with the LV coal type, while HVoperation resulted in a smooth Femet texture. Examples of typical straggling andsmooth Femettextures are presented in Figure 33.

Figure 33 Typical examples of a straggling Femettexture (left)and a smooth Femettexture (right).

Rounded and assembled slag areas were identified during HV coal operation, whileLV coal injection resulted in scattered pattern of slag. The test conditions in theEBF, in terms of reduction potential of the gas and in-furnace temperature profile,did not result in formation of distinct product layers in the reduced pellets.

In the BF simulation tests the temperature and reduction conditions were the samefor all the pellets tested. From this point of view it was reasonable that differencesin the pellet texture did not occur between pellets of the same type. It should be

noted that the influence of original pellet texture has not been considered in any ofthe investigations.

4.3 Process

Softened areas of slag and smoother Femet texture in the pellets indicated thatinjection of the HV coal type increased the shaft temperature in the EBF. Due tothe relatively high amount of volatile matters in the coal, the H2 content in theascending gas was higher during injection of the HV coal compared to during



60/124

48

injection of the LV coal. Increasing the PCR increases the H2content of the gas,independent of the coal type. The differences in CO were quite small.Consequently, the changes in reduction potential under different process conditions

seemed to be more or less dependent on the H2content.

Fe3C was observed in pellets taken out with the lower shaft probe in the EBF. Thechemical analyses showed an increase in C content during injection of the HV coaltype. It was therefore concluded that the formation of Fe3C was more frequentduring injection of the HV coal type. It was difficult to determine the cause ofincreased Fe3C formation during injection of the HV coal, but there are some

possible explanations. Fe3C was formed in the shaft between the positions of theshaft probes. A higher amount of H2in the ascending gas facilitated the iron oxide

reduction by H2 and, as a consequence the CO concentration available forcarburization increased as the CO to a lesser extent contributed to the reduction.Reaction of Fe3C with wustite, reaction 14, was more likely to occur duringinjection of the LV coal type, since the pellet reduction degree was lower.Therefore, a lower carbon content should be expected when using the LV coalwhich is in agreement with the present results.

The differences in hot metal Si at SSAB BF No. 3 can not be correlated to the ironore raw materials used during the sampling period. Differences in FeO contentinfluencing the hot metal Si content were not expected in the EBF, since nosignificant differences in average reduction degrees in pellets taken out with thelower shaft probe were observed. During the EBF test, the trend of the hot metalwas decreasing Si content, mainly due to increased CaO/SiO2of the slag.

4.4 Generation of Fines

The highest amount of fine material was observed during the HV coal operationduring oxygen supply to the coal lance, which could be seen in the 0-6.3 mmmaterial fraction taken out with the upper shaft probe. When the share of the 0-0.5mm fraction increased, the analysis of corresponding samples showed an increasedamount of Fe at the same time as the C content decreased. Injection of the HV coalat the high PCR increased the Fe content in the fines in the material taken out withthe upper shaft probe. The generation of fine material leaving the EBF, and Fecontent in flue dust, increased when injecting the HV coal. The highest pelletstrength was observed at the highest reduction degrees; however, an increased



61/124

49

generation of fines in the 0-0.5 mm fraction was also evident, resulting in higherFemetlosses through the top gas.

It is believed that the Femettextures formed in the pellet periphery during reductionwill affect the generation of fines. During abrasion, the edges of the stragglingFemet areas get stuck to each other, while a smooth Femet surface area does notexhibit this property. Since the Femetareas with smooth Femetsurfaces dominatedduring HV coal injection, the Femet texture is probably one explanation for theincrease in fines generation compared to LV coal operation.

4.5 Sources of Error

According to the literature, a random distribution of pores has been observed inmagnetite formed during reduction of hematite pellets, but magnetite withoutinternal porosity was also observed.[18] The magnetite textures were obtained atdifferent temperatures and at different reduction gas compositions. In the presentinvestigations it was assumed that no internal porosity existed in the magnetite andthat the areas of similar colour, in which internal porosity was observed, consistedof wustite. There is a possibility that magnetite and wustite were not kept separateat all times.

Iron oxide reduction and carbon deposition occur simultaneously.[2]

It cannot beconcluded for certain that carbon deposition did not occur during the laboratoryreduction tests performed and if so was the case, a higher reduction degreecompared to the ones presented might have been reached.

4.6 Concluding Discussion

The results from the EBF tests indicated that the test conditions in terms of thereduction potential of the gas, temperature profile and time for all cases resulted in

an extension of the thermal reserve zone that was sufficient for reduction of pelletsunder all the investigated process conditions. The different Femettextures observedin pellets taken out with the lower shaft probe indicated differences in reductionconditions at positions above the lower shaft probe in the EBF. The results indicatethat operation with an LV coal was preferred if a high PCR was desired. Thegeneration of Femet fines would most likely have been decreased, thanks to theformation of straggling Femetgrains. The initial reduction conditions, in terms oftemperature and gas composition, were in the performed investigations conclusivefor the pellet texture formed.



62/124

50



63/124


64/124

52



65/124

53

6 FUTURE WORKThe performed investigation has shown that the initial pellet reduction conditions,

in terms of temperature and reduction gas composition, were decisive for thetexture in the pellet periphery. An investigation should be carried out in order todetermine the relationship between the Femettexture in the pellet periphery and thegeneration of Femetfines. An increased understanding in this matter would facilitatethe possibility to predict fines generation from pellets as an outcome of, forexample, different PCR. Investigations in order to assess whether there is acorrelation between top gas temperature and composition and generation of and Fecontent in flue dust and sludge should be carried out.

An increased PCR increases the ore-to-coke ratio. Higher demands are placed onpellet strength and ability to withstand generation of Femet fines. Investigation topredict the highest PCR at which a normal amount of fines is generated ought to becarried out with different injection coal types or different reducing agents used forinjection. At the same time, the pellet strength should be investigated.

Analyses of pellets taken out with the lower shaft probe in the EBF showedcontents of Fe3C. Carbon deposition and carburization take place in reduction gasmixtures of H

2and CO.[2]It is of interest to know more about how the reduction

gas composition and temperature, which are affected by among other things thePCR, influence the carburization of pellets and their metallurgical properties. It isalso important to learn more about how carburization of pellets influences the

pellet strength and formation of Femet fines. Reduction experiments, followed bypellet characterization, should be carried out in laboratory scale for differenttemperatures and reduction gas composition profiles.



66/124

54



67/124

55

7 REFERENCES1 A.K. Biswas, Principles of Blast Furnace Ironmaking, Cootha Publishing

House, Australia, 1981.2 N. Towhidi and J. Szekely. The Influence of Carbon Deposition on theReduction Kinetics of Commercial Grade Hematite Pellets with CO, H2and

N2, Metallurgical Transactions B, Vol 14B, September 1983, pp. 359-367.3 E. T. Turkdogan and J. V. Vinters. Catalytic Effect of Iron on Decomposition

of Carbon Monoxide: I. Carbon Deposition in H2-CO Mixtures,Metallurgical Transactions, Vol. 5, 1974, pp. 11-19.

4 A. A. El-Geassy and I. Nasr. Influence of Original Structure on the Kineticsand Mechanisms of Carbon Monoxide Reduction of Hematite Compacts,

ISIJ International, Vol. 30, No. 6, 1990, pp. 417-425.5 M. I. Nasr, A. A. Omar, M. M. Messien and A. A. El-Geassy. Carbon

Monoxide Reduction and Accompanying Swelling of Iron Oxide Compacts,ISIJ International, Vol. 36, No. 2, 1996, pp. 164-171.

6 K. Takeda. Reaction and Transport Phenomena Relevant to the Si Transfer atthe Lower Part of the Blast Furnace, High Temperature Materials and

Processes, Vol. 19, No. 3-4, 2000, pp. 177-186.7 T. Miwa, X. Tong, J. B. Guillot and A. Rist. A Laboratory Study of Silicon

Transfer Under Blast Furnace Conditions, European Ironmaking Congress,Aachen, 1986.

8 Y. Matsui, S. Mori and F. Noma, Kinetics of Silicon Transfer fromPulverized Coal Injection into Blast Furnace under Intensive Coal Injection,

ISIJ International, Vol. 43, No. 7, 2003, pp. 997-1002.9 A. Maki, A. Sakai, N. Takagaki, K. Mori, T. Ariyama, M. Sato and R. Murai.

High Rate Coal Injection of 218 kg/t at Fukuyama No. 4 Blast Furnace,ISIJInternational, Vol. 36, No. 6, 1996, pp. 650-657.

10 S-H. Yi, W-W Huh, C-H. Rhee and B-R. Cho, Softening and melting

properties of pellets for a high level of pulverized coal-injected blast furnaceoperation, Scandinavian Journal of Metallurgy, Vol. 28, No. 6, December1999, pp. 260-265.

11 K-H. Peters, M. Peters, B. Korthas, K. Mlheims and K. Kreibich. Limits ofcoal injection, Metallurgical Plant and Technology International, No. 6,1990, pp. 32-43.

12 K. Yamaguchi, H. Ueno, S. Matsunaga, K. Kakiuchi and S. Amano. Test onHigh-rate Pulverized Coal Injection Operation at Kimitsu No. 3 BlastFurnace,ISIJ International, Vol. 35, No. 2, 1995, pp. 148-155.



68/124


69/124

57

25 S. Hayashi and Y. Iguchi. Effect of Coexistence of Potassium and Sulphur onAbnormal Swelling during Reduction of Hematite Pellets,ISIJ International,Vol. 29, No. 8, 1989, pp. 642-649.

26 LKAB Products 200427 L. Sundqvist kvist, A. Dahlstedt, M. Hallin. The Effect on Blast furnaceProcess of changed Pellet Size as a Result of segregation in Raw MaterialHandling,ISS 32ndIronmaking Conference, Baltimore, USA, 2001.

28 A. Dahlstedt, M. Hallin, M. Tottie, LKABs Experimental Blast Furnace forEvaluation of Iron Ore Products, Scanmet 1, Lule Sweden, 1999.

29 LKAB, private conversation30 P.L. Hooey, J. Sterneland, M. Hallin, Evaluation of High Temperature

Properties of Blast Furnace Burden, 1stInternational Meeting on Ironmaking,

Belo Horizonte Brazil, September 2001.



70/124



71/124

Appendix I

Correlation Between Pellet Reduction and Some BlastFurnace Operation Parameters

U. Leimalm, L. Sundqvist kvist, A. Brnnmark and B. Bjrkman

AISTech 2006 Proceedings of the Iron & Steel TechnologyConference, Cleveland, USA, May 2006



72/124



73/124

Correlation Between Pellet Reduction and Some Blast Furnace Operation Parameters

Ulrika Leimalm*, Lena Sundqvist-kvist**, Anna Brnnmark***, Bo Bjrkman**Lule University of Technology

SE-971 87 LuleSweden

Tel: +46-920-491000Fax: +46-920-491199

E-mail: [email protected],bo.bjorkman@lt

Documents

Pellet Reduction Properties Under Different BF Operating Conditions